Autofocus method in digital holographic microscopy

The determination of focused image plane is the key of numerical reconstruction of wavefront. In this paper, three autofocus algorithms for digital holographic microscopy, including statistical algorithm, histogram-based algorithms and Fourier spectrum algorithm, are studied based on experimental investigation. For pure amplitude objects, three autofocusing evaluation functions for digital holographic microscopy are compared and analyzed based on Fresnel and angular spectrum algorithms, including unimodality, sharpness, veracity, distance range and computing time. The results demonstrate that Fresnel transform algorithm can be absolutely used to digital holographic autofocusing. When the focus distance is reached, the value of the focused evaluation function is maximum for pure amplitude object. For real digital hologram, there are better unimodality near the focused image plane for all three autofocusing evaluation functions, and the same focused position is obtained by these algorithms. Variance algorithm has better sharpness then others. Fourier spectrum algorithm is the most time-effective one, which is the optimal one in digital holographic microscopy. Moreover, the focusing computation time can be decreased dramatically by choosing part of the reconstructed image as the focusing area.